Skip to content

Instantly share code, notes, and snippets.

@hungson175
hungson175 / workspace-mcp-403-diagnosis.md
Created June 3, 2026 08:44
Google Workspace MCP (Developer Preview) — 403 'caller does not have permission' diagnosis & fix

Google Workspace MCP (Developer Preview) — 403 diagnosis & fix

Account under test: son.pham9@mservice.com.vn GCP project (owns the OAuth client): momo-workspace-mcp-preview (project # 193921716285) OAuth client (Desktop): 193921716285-8fh0o97ipqbve0dg5quubj29gklggqdg.apps.googleusercontent.com (client secret intentionally omitted — never share it publicly) Guide: https://developers.google.com/workspace/guides/configure-mcp-servers


@hungson175
hungson175 / enable-workspace-mcp-apis.md
Last active June 3, 2026 07:04
Enable Google Workspace MCP APIs on momo-workspace-mcp-preview

Enable Google Workspace MCP APIs — momo-workspace-mcp-preview

Project: momo-workspace-mcp-preview (project number 193921716285) From: son.pham9@mservice.com.vn Context: Following Google's official guide — https://developers.google.com/workspace/guides/configure-mcp-servers The OAuth Desktop client you gave me is configured and the MCP client connects fine. The only remaining blocker is that the hosted MCP service APIs are not enabled on the project. I don't have admin access to this project, so could you please enable them?

⏱️ CURRENT STATUS (what's left)

Confirmed enabled: gmailmcp + gmail, drivemcp + drive, calendarmcp + calendar-json. Thank you!

@hungson175
hungson175 / llm_coding_cost_20260527.md
Created May 27, 2026 15:28
LLM Cost for Coding Agents — May 2026 (effective price + AA Intelligence Index)

LLM Cost for Coding Agents — May 2026

Cached input price is everything. A coding agent's token mix is 95.6% cache reads (measured: 8.04B tokens, May 2026). Headline input price is nearly irrelevant.

Formula: Effective $/1M = cache_read × 0.956 + cache_write × 0.027 + input × 0.013 + output × 0.003


# Model Cache Read Input Output Cache Write Effective $/1M vs cheapest AA Intelligence Index
@hungson175
hungson175 / macbook-hcmc-buyers-guide-20260527.md
Last active May 27, 2026 14:44
MacBook M2/M3/M4/M5 + 24/32GB RAM — HCMC buyer's guide, new + MDM, live-verified 27 May 2026

MacBook Buyer's Guide — HCMC (27 May 2026)

Target: M2 / M3 / M4 / M5 MacBook Pro 14" or MacBook Air, 24GB or 32GB RAM. Trusted-dealer-first list. All prices VND, verified live the day of publication. Best deals at the top.


⚡ Headline deals (trusted shops, verified today)

🥇 MBA 13" M4 24GB/512GB MDM — 23,900,000₫ — PTStore website

macpt.vn product page

@hungson175
hungson175 / 2026-05-16-gemma4-vs-qwen3.6-real-coding.md
Created May 16, 2026 04:56
Gemma 4 26B A4B vs Qwen3.6-27B vs Qwen3.6-35B-A3B — Real-Coding Verdict (May 2026)

Gemma 4 26B A4B vs Qwen3.6-27B Dense vs Qwen3.6-35B-A3B — Real-Coding Verdict

Research date: 16 May 2026 Constraint: Real coding agent work (NOT competitive programming). Excludes LiveCodeBench, Codeforces, HumanEval-only. Hardware constraint: Single 24GB GPU (RTX 3090Ti)


TL;DR

@hungson175
hungson175 / kg-field-manual-sonph-20260515.md
Created May 15, 2026 16:03
Long-Term Memory Is the Self-Improvement Mechanism: A Field Manual for SonPH's Agent Army v2 - May 2026

Long-Term Memory Is the Self-Improvement Mechanism: A Field Manual for SonPH's Agent Army

May 15, 2026 — Ho Chi Minh City


The Realization You Already Had

You reverse-engineered Claude Code and identified three things that make it work:

@hungson175
hungson175 / kg-field-manual-sonph-20260515.md
Created May 15, 2026 15:55
Long-Term Memory Is the Self-Improvement Mechanism: A Field Manual for SonPH's Agent Army - May 2026

Long-Term Memory Is the Self-Improvement Mechanism: A Field Manual for SonPH's Agent Army

May 15, 2026 — Ho Chi Minh City


The Realization You Already Had

You reverse-engineered Claude Code and identified three things that make it work:

@hungson175
hungson175 / MOAT-Eval-Report-v10-CTO-discussion-2026-05-09.md
Last active May 9, 2026 14:21
MOAT Eval Report v10 — CTO discussion branch 2026-05-09

MOAT Data Agent — Evaluation Process Design

Version: v10 | Prepared by: Son Pham | Date: 2026-05-08


Executive Summary

MOAT is MoMo's AI data agent that answers business questions in natural language against BigQuery. This document describes the evaluation framework — both how we measure quality (IFAB) and how we continuously improve without degrading what already works.

Key points:

@hungson175
hungson175 / Speaker-Diarization-API-Research-20260509.md
Created May 9, 2026 03:49
Speaker Diarization API Research — May 2026 (for Meeting Note app)
title Speaker Diarization for Meeting Note App — Best Options May 2026
date 2026-05-09
type research
status inbox
project meeting-note-app
related soniox-stt-service.md, soniox-context-injection.md

Speaker Diarization for Meeting Note App — May 2026

@hungson175
hungson175 / SKILL.md
Created May 8, 2026 11:56
use-my-pkm Claude Code skill (place at ~/.claude/skills/use-my-pkm/SKILL.md)
name use-my-pkm
description Access and work with the user's personal knowledge management vault using $MY_PKM_PATH environment variable. Use when the user asks to work with their PKM system, notes, Kanban board, inbox items, or knowledge vault. Supports reading/writing notes, managing Kanban tasks, organizing inbox, navigating the vault structure via progressive disclosure, and semantic search via qmd (BM25 + vector + reranking). Also use when the user asks to search, find, or look up notes by meaning/concept.
private true

Use My PKM — LLM Wiki

Work with Boss's personal knowledge wiki (LLM Wiki pattern, Karpathy-style).